In the late 1980s, the UK's Overseas Development Administration – now the Department for International Development (DFID) – contracted out the management of much of its centrally funded research. Indeed, paragraph 35 of its written evidence presented to the House of Commons Select Committee on Science and Technology on 12 January 2004 mentions 14 health programmes, 10 programmes in renewable natural resources, and 8 for social science.

The strategic development of these programmes has been left largely with outsourced programme managers and, notwithstanding the constraint of fixed-price contracts and a limited mandate, we already do much of what DFID is aspiring to do in the future and which David Dickson recommends in his editorial (see Britain must do more to link science to development).

The new research strategy of DFID, for which publication is imminent, does not draw upon these diverse experiments in research management and the promotion of pro-poor research results. Two improvements are especially necessary in the new strategy if DFID's research is to contribute to attainment of the Millennium Development Goals.

The first is to manage the research-development-application continuum as a whole. At present, the continuum is broken so that different agencies are responsible for funding strategic (multi-country) research, its local adaptation, the piloting of applications, and the wider uptake of outputs.

The second improvement is a parallel and perennial exercise to build national research capacity in developing countries. At present, many of their agencies for research in renewable natural resources (agriculture, fisheries, livestock, and forestry) are weak in status, have inappropriate mandates, intermittent training and insufficient operational funds.

These problems can be ascribed in part to earlier weaknesses in the education systems. Spasmodic institutional strengthening, linked to short-term research projects, helps only the lucky few associated with, for example, DFID-funded short projects. This is only pecking at the problem, not solving it. Not a spoonful of cough mixture, but a full course of iron tonic is needed.